Choosing the right model
Start with "Default", switch to a stronger model for difficult or long tasks and to an economical one for simple bulk work.
Most of the time you do not need to pick a model. This page helps in the cases where the choice makes a difference to quality, credits or data protection. How the model picker works is in Choosing a model.
Start with "Default"
The first entry in the model picker, Default, stands for your personal Default model, otherwise your group's, otherwise the workspace's. Your admin chose it to suit the region, the budget and typical tasks. A faster model may answer short questions. Switch only when you have a reason.

The card next to each model shows its Credit usage (Low, Medium, High), where it runs and whether it can read images and use tools.
Which task, which model
| Task | Choice | Why |
|---|---|---|
| Emails, short texts, quick questions | "Default" | Enough for everyday work and saves credits. |
| Sorting, rewording or categorising many short texts | Credit usage Low | Simple tasks do not need a large model. |
| Reviewing a contract, multi-step analysis, delicate wording | Credit usage High | More care with details and instructions. |
| Very long documents or many attachments | A model with a large context | Otherwise not everything fits in one request. |
| Reading screenshots, photos, diagrams | A model without the note Cannot read images. | Other models do not see the image. |
| Brain, web search, connectors, files, company templates | A model without the note No web or brain search, and it creates no files. | Without tools these features are missing. |
| Client or personnel data | Security mode | Your admin sets the model, you do not choose. |
When a stronger model pays off
When a mistake would be costly or "Default" visibly reaches its limits:
- The answer misses exceptions, deadlines or conditions that are in the document.
- The model does not follow clear instructions on format or length.
- A calculation or line of reasoning with several steps goes wrong along the way.
How to test it: pick a stronger model and regenerate the last answer with it. Custodos keeps both versions. If the difference matters for this kind of task, use that model for it from now on. If you need it almost always, set it as your personal Default model, see Personal settings.
Models that think before answering do better at difficult analyses, but use more credits than the length of the answer suggests. For simple questions a model without thinking is enough.
Long documents
When a chat gets long, Custodos summarises older messages; attached files stay available. If a single message with its attachments is too large, remove an attachment, start a new chat or switch to a model with a larger context. If you regularly work with many documents on the same topic, a brain is better: it gives the model only the relevant passages.
When the budget is used up
From then until the reset you keep working with the economy model. Model choice, web search, brain, files and images are paused until then, and nothing is billed on top. See Budgets and limits.
Frequently asked questions
Your admin has not enabled it, it runs outside your workspace's processing region, or it is enabled only for certain groups. If a group keeps your data in Switzerland, you see only models that process there. See Models.
Either your budget is used up, or the chat runs in security mode, which sets the model. A note at the message box says which case applies.